Import Packages¶
In [1]:
import matplotlib.pyplot as plt
import matplotlib.image as mpimg
import numpy as np
import cv2
import glob
import ntpath
%matplotlib inline
Calibrate Camera¶
Calibration Images¶
In [2]:
cal_images = sorted(glob.glob('camera_cal/*.jpg'))
fig = plt.figure(figsize = (20, 20))
for i in range(len(cal_images)):
cal_image = cal_images[i]
image = mpimg.imread(cal_image)
ax = fig.add_subplot(5,4,i+1)
ax.set_title(cal_image)
ax.imshow(image.squeeze())
Find Corners¶
In [3]:
object_points = []
image_points = []
# Array to store object points and image points from all the calibration images
objp = np.zeros((6*9, 3), np.float32)
objp[:,:2] = np.mgrid[0:9, 0:6].T.reshape(-1, 2)
fig = plt.figure(figsize = (20, 20))
for i in range(len(cal_images)):
cal_image = cal_images[i]
image = mpimg.imread(cal_image)
gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
ret, corners = cv2.findChessboardCorners(gray, (9, 6), None)
if ret == True:
image_points.append(corners)
object_points.append(objp)
corner_image = cv2.drawChessboardCorners(image, (9, 6), corners, ret)
ax = fig.add_subplot(5,4,i+1)
ax.set_title(cal_image)
ax.imshow(corner_image)
else:
ax = fig.add_subplot(5,4,i+1)
ax.set_title(cal_image + ': cannot find corners')
ax.imshow(image)
Undistort Images¶
In [4]:
camera_mtx = None
camera_dist = None
def cal_undistort(img, objpoints, imgpoints):
# Use cv2.calibrateCamera() and cv2.undistort()
gray = cv2.cvtColor(img,cv2.COLOR_BGR2GRAY)
ret, mtx, dist, rvecs, tvecs = cv2.calibrateCamera(objpoints, imgpoints, gray.shape[::-1], None, None)
camera_mtx = mtx
camera_dist = dist
undist = cv2.undistort(img, mtx, dist, None, mtx)
return undist
fig = plt.figure(figsize = (20, 100))
for i in range(len(cal_images)):
cal_image = cal_images[i]
image = mpimg.imread(cal_image)
undist = cal_undistort(image, object_points, image_points)
ax1 = fig.add_subplot(20,2,(i*2)+1)
ax1.set_title(cal_image + ': Original')
ax1.imshow(image)
ax2 = fig.add_subplot(20,2,(i*2)+2)
ax2.set_title('Undistorted')
ax2.imshow(undist)
Test Images¶
In [5]:
test_images = sorted(glob.glob('test_images/*.jpg'))
fig = plt.figure(figsize = (20, 50))
for i in range(len(test_images)):
test_image = test_images[i]
image = mpimg.imread(test_image)
undist = cal_undistort(image, object_points, image_points)
ax1 = fig.add_subplot(8,2,(i*2)+1)
ax1.set_title(ntpath.basename(test_image))
ax1.imshow(image.squeeze())
ax2 = fig.add_subplot(8,2,(i*2)+2)
ax2.set_title('Undistorted')
ax2.imshow(undist)
Color Transform and Gradients¶
In [6]:
def binary(img):
# Convert to HLS color space and separate the S channel
# Note: img is the undistorted image
hls = cv2.cvtColor(img, cv2.COLOR_RGB2HLS)
s_channel = hls[:,:,2]
# Grayscale image
# NOTE: we already saw that standard grayscaling lost color information for the lane lines
# Explore gradients in other colors spaces / color channels to see what might work better
gray = cv2.cvtColor(img, cv2.COLOR_RGB2GRAY)
# Sobel x
sobelx = cv2.Sobel(gray, cv2.CV_64F, 1, 0) # Take the derivative in x
abs_sobelx = np.absolute(sobelx) # Absolute x derivative to accentuate lines away from horizontal
scaled_sobel = np.uint8(255*abs_sobelx/np.max(abs_sobelx))
# Threshold x gradient
thresh_min = 20
thresh_max = 100
sxbinary = np.zeros_like(scaled_sobel)
sxbinary[(scaled_sobel >= thresh_min) & (scaled_sobel <= thresh_max)] = 1
# Threshold color channel
s_thresh_min = 170
s_thresh_max = 255
s_binary = np.zeros_like(s_channel)
s_binary[(s_channel >= s_thresh_min) & (s_channel <= s_thresh_max)] = 1
# Stack each channel to view their individual contributions in green and blue respectively
# This returns a stack of the two binary images, whose components you can see as different colors
color_binary = np.dstack(( np.zeros_like(sxbinary), sxbinary, s_binary)) * 255
# Combine the two binary thresholds
combined_binary = np.zeros_like(sxbinary)
combined_binary[(s_binary == 1) | (sxbinary == 1)] = 1
return combined_binary
fig = plt.figure(figsize = (20, 50))
for i in range(len(test_images)):
test_image = test_images[i]
image = mpimg.imread(test_image)
undist = cal_undistort(image, object_points, image_points)
image_binary = binary(undist)
ax1 = fig.add_subplot(8,2,(i*2)+1)
ax1.set_title(ntpath.basename(test_image) + ': Undistorted')
ax1.imshow(undist)
ax2 = fig.add_subplot(8,2,(i*2)+2)
ax2.set_title('Combined S channel and gradient thresholds')
ax2.imshow(image_binary, cmap='gray')
Perspective Transform¶
In [7]:
warp_src = np.float32([[595, 450],
[690, 450],
[1040, 670],
[270, 670]])
warp_dst = np.float32([[250,0],
[1030,0],
[1030,720],
[250,720]])
M = cv2.getPerspectiveTransform(warp_src, warp_dst)
M_inv = cv2.getPerspectiveTransform(warp_dst, warp_src)
def warp(img):
img_size = img.shape[1::-1]
warped = cv2.warpPerspective(img, M, img_size)
return warped
def draw_region(img):
cv2.polylines(img,[np.array(warp_src, np.int32).reshape((-1,1,2))],True,(255,0,0), thickness = 2)
fig = plt.figure(figsize = (20, 100))
for i in range(len(test_images)):
test_image = test_images[i]
image = mpimg.imread(test_image)
undist = cal_undistort(image, object_points, image_points)
warp_image = warp(undist)
image_binary = binary(undist)
warp_binary = warp(image_binary)
draw_region(undist)
ax1 = fig.add_subplot(16,2,(i*4)+1)
ax1.set_title(ntpath.basename(test_image) + ': Undistorted')
ax1.imshow(undist)
ax2 = fig.add_subplot(16,2,(i*4)+2)
ax2.set_title('Perspective Transformed')
ax2.imshow(warp_image, cmap='gray')
ax3 = fig.add_subplot(16,2,(i*4)+3)
ax3.set_title('Color Transformed')
ax3.imshow(image_binary, cmap='gray')
ax4 = fig.add_subplot(16,2,(i*4)+4)
ax4.set_title('Perspective Transformed')
ax4.imshow(warp_binary, cmap='gray')
Finding the Lane¶
Histogram¶
In [8]:
def histogram(img):
img_histogram = np.sum(img[img.shape[0]//2:,:], axis=0)
return img_histogram
fig = plt.figure(figsize = (20, 50))
for i in range(len(test_images)):
test_image = test_images[i]
image = mpimg.imread(test_image)
undist = cal_undistort(image, object_points, image_points)
image_binary = binary(undist)
warp_binary = warp(image_binary)
image_histogram = histogram(warp_binary)
ax1 = fig.add_subplot(8,2,(i*2)+1)
ax1.set_title(ntpath.basename(test_image))
ax1.imshow(warp_binary, cmap='gray')
ax2 = fig.add_subplot(8,2,(i*2)+2)
ax2.set_title('Histogram')
ax2.plot(image_histogram)
Helper Class¶
In [9]:
# Define a class to receive the characteristics of each line detection
class Line():
def __init__(self):
# was the line detected in the last iteration?
self.detected = True
#polynomial coefficients for the most recent fit
self.current_fit = [np.array([False])]
#radius of curvature of the line in some units
self.radius_of_curvature = None
#distance in meters of vehicle center from the line
self.line_base_pos = None
#x values for detected line pixels
self.allx = None
#y values for detected line pixels
self.ally = None
Find and Highlight Lane Lines¶
In [10]:
def detect_lines_inner(img, left_line_inds, right_line_inds, nonzerox, nonzeroy, margin):
left_line = Line()
right_line = Line()
# Again, extract left and right line pixel positions
left_line.allx = nonzerox[left_line_inds]
left_line.ally = nonzeroy[left_line_inds]
right_line.allx = nonzerox[right_line_inds]
right_line.ally = nonzeroy[right_line_inds]
# Fit a second order polynomial to each
left_line.current_fit = np.polyfit(left_line.ally, left_line.allx, 2)
right_line.current_fit = np.polyfit(right_line.ally, right_line.allx, 2)
# Generate x and y values for plotting
ploty = np.linspace(0, img.shape[0]-1, img.shape[0] )
left_fitx = left_line.current_fit[0]*ploty**2 + left_line.current_fit[1]*ploty + left_line.current_fit[2]
right_fitx = right_line.current_fit[0]*ploty**2 + right_line.current_fit[1]*ploty + right_line.current_fit[2]
left_line_window1 = np.array([np.transpose(np.vstack([left_fitx-margin, ploty]))])
left_line_window2 = np.array([np.flipud(np.transpose(np.vstack([left_fitx+margin,
ploty])))])
left_line_pts = np.hstack((left_line_window1, left_line_window2))
right_line_window1 = np.array([np.transpose(np.vstack([right_fitx-margin, ploty]))])
right_line_window2 = np.array([np.flipud(np.transpose(np.vstack([right_fitx+margin,
ploty])))])
right_line_pts = np.hstack((right_line_window1, right_line_window2))
out_img = np.dstack((img, img, img))*255
out_img[nonzeroy[left_line_inds], nonzerox[left_line_inds]] = [255, 0, 0]
out_img[nonzeroy[right_line_inds], nonzerox[right_line_inds]] = [0, 0, 255]
# Create an output image to draw on and visualize the result
window_img = np.zeros_like(out_img)
cv2.fillPoly(window_img, np.int_([left_line_pts]), (0,255, 0))
cv2.fillPoly(window_img, np.int_([right_line_pts]), (0,255, 0))
result = cv2.addWeighted(out_img, 1, window_img, 0.3, 0)
return result, left_line, right_line
def detect_lines(img, highlight = True):
margin = 100
image_histogram = histogram(img)
# Find the peak of the left and right halves of the histogram
# These will be the starting point for the left and right lines
midpoint = np.int(image_histogram.shape[0]/2)
leftx_base = np.argmax(image_histogram[:midpoint])
rightx_base = np.argmax(image_histogram[midpoint:]) + midpoint
# Choose the number of sliding windows
nwindows = 9
# Set height of windows
window_height = np.int(img.shape[0]/nwindows)
# Identify the x and y positions of all nonzero pixels in the image
nonzero = img.nonzero()
nonzeroy = np.array(nonzero[0])
nonzerox = np.array(nonzero[1])
# Current positions to be updated for each window
leftx_current = leftx_base
rightx_current = rightx_base
# Set minimum number of pixels found to recenter window
minpix = 50
# Create empty lists to receive left and right line pixel indices
left_line_inds = []
right_line_inds = []
# Step through the windows one by one
for window in range(nwindows):
# Identify window boundaries in x and y (and right and left)
win_y_low = img.shape[0] - (window+1)*window_height
win_y_high = img.shape[0] - window*window_height
win_xleft_low = leftx_current - margin
win_xleft_high = leftx_current + margin
win_xright_low = rightx_current - margin
win_xright_high = rightx_current + margin
# Identify the nonzero pixels in x and y within the window
good_left_inds = ((nonzeroy >= win_y_low) & (nonzeroy < win_y_high) &
(nonzerox >= win_xleft_low) & (nonzerox < win_xleft_high)).nonzero()[0]
good_right_inds = ((nonzeroy >= win_y_low) & (nonzeroy < win_y_high) &
(nonzerox >= win_xright_low) & (nonzerox < win_xright_high)).nonzero()[0]
# Append these indices to the lists
left_line_inds.append(good_left_inds)
right_line_inds.append(good_right_inds)
# If you found > minpix pixels, recenter next window on their mean position
if len(good_left_inds) > minpix:
leftx_current = np.int(np.mean(nonzerox[good_left_inds]))
if len(good_right_inds) > minpix:
rightx_current = np.int(np.mean(nonzerox[good_right_inds]))
# Concatenate the arrays of indices
left_line_inds = np.concatenate(left_line_inds)
right_line_inds = np.concatenate(right_line_inds)
return detect_lines_inner(img, left_line_inds, right_line_inds, nonzerox, nonzeroy, margin)
fig = plt.figure(figsize = (20, 50))
for i in range(len(test_images)):
test_image = test_images[i]
image = mpimg.imread(test_image)
undist = cal_undistort(image, object_points, image_points)
image_binary = binary(undist)
warp_binary = warp(image_binary)
lines_highlighted, left_line, right_line = detect_lines(warp_binary)
ax1 = fig.add_subplot(8,2,(i*2)+1)
ax1.set_title(ntpath.basename(test_image))
ax1.imshow(warp_binary, cmap='gray')
ax2 = fig.add_subplot(8,2,(i*2)+2)
ax2.set_title('Lines Highlighted')
ax2.imshow(lines_highlighted)
Calculating Curvature¶
In [11]:
def cal_curvature_and_pos(left_line, right_line):
# Define y-value where we want radius of curvature
# I'll choose the maximum y-value, corresponding to the bottom of the image
y_eval = 720
width = 1280
# Define conversions in x and y from pixels space to meters
ym_per_pix = 30/720 # meters per pixel in y dimension
xm_per_pix = 3.7/700 # meters per pixel in x dimension
# Fit new polynomials to x,y in world space
left_fit_cr = np.polyfit(left_line.ally*ym_per_pix, left_line.allx*xm_per_pix, 2)
right_fit_cr = np.polyfit(right_line.ally*ym_per_pix, right_line.allx*xm_per_pix, 2)
# Calculate the new radii of curvature
left_line.radius_of_curvature = ((1 + (2*left_fit_cr[0]*y_eval*ym_per_pix + left_fit_cr[1])**2)**1.5) / np.absolute(2*left_fit_cr[0])
right_line.radius_of_curvature = ((1 + (2*right_fit_cr[0]*y_eval*ym_per_pix + right_fit_cr[1])**2)**1.5) / np.absolute(2*right_fit_cr[0])
# Now our radius of curvature is in meters
left_bottom_x = left_line.current_fit[0]*y_eval**2 + left_line.current_fit[1]*y_eval + left_line.current_fit[2]
right_bottom_x = right_line.current_fit[0]*y_eval**2 + right_line.current_fit[1]*y_eval + right_line.current_fit[2]
# Line position respective to center of image
left_line.line_base_pos = (left_bottom_x - (width/2)) * xm_per_pix
right_line.line_base_pos = (right_bottom_x - (width/2)) * xm_per_pix
fig = plt.figure(figsize = (20, 25))
for i in range(len(test_images)):
test_image = test_images[i]
image = mpimg.imread(test_image)
undist = cal_undistort(image, object_points, image_points)
image_binary = binary(undist)
warp_binary = warp(image_binary)
lines_highlighted, left_line, right_line = detect_lines(warp_binary)
cal_curvature_and_pos(left_line, right_line)
curverad = round((left_line.radius_of_curvature + right_line.radius_of_curvature) / 2)
curverad_str = 'Curve radius: ' + str(curverad) + 'm'
lane_base_pos = round((left_line.line_base_pos + right_line.line_base_pos) / 2, 4)
if lane_base_pos > 0:
position_str = str(lane_base_pos * 100) + 'cm right of center'
elif lane_base_pos < 0:
position_str = str(lane_base_pos * -100) + 'cm left of center'
else:
position_str = 'right at center'
ax = fig.add_subplot(4,2,i+1)
ax.set_title(ntpath.basename(test_image) + ': ' + curverad_str + ', ' + position_str)
ax.imshow(lines_highlighted)
Detect Lines Starting from Previous Lines¶
In [12]:
def detect_lines_from_previous_lines(img, prev_left_line, prev_right_line):
nonzero = img.nonzero()
nonzeroy = np.array(nonzero[0])
nonzerox = np.array(nonzero[1])
margin = 100
left_line_inds = ((nonzerox > (prev_left_line.current_fit[0]*(nonzeroy**2) + prev_left_line.current_fit[1]*nonzeroy +
prev_left_line.current_fit[2] - margin)) & (nonzerox < (prev_left_line.current_fit[0]*(nonzeroy**2) +
prev_left_line.current_fit[1]*nonzeroy + prev_left_line.current_fit[2] + margin)))
right_line_inds = ((nonzerox > (prev_right_line.current_fit[0]*(nonzeroy**2) + prev_right_line.current_fit[1]*nonzeroy +
prev_right_line.current_fit[2] - margin)) & (nonzerox < (prev_right_line.current_fit[0]*(nonzeroy**2) +
prev_right_line.current_fit[1]*nonzeroy + prev_right_line.current_fit[2] + margin)))
return detect_lines_inner(img, left_line_inds, right_line_inds, nonzerox, nonzeroy, margin)
Draw Lane¶
In [13]:
left_line = None
right_line = None
def draw_lane(img):
global left_line
global right_line
prev_left_line = None
prev_right_line = None
y_eval = 720
undist = cal_undistort(img, object_points, image_points)
warp_image = warp(undist)
image_binary = binary(undist)
warp_binary = warp(image_binary)
if left_line == None or right_line == None or left_line.detected or not right_line.detected:
lines_highlighted, left_line, right_line = detect_lines(warp_binary)
else:
prev_left_line, prev_right_line = left_line, right_line
lines_highlighted, left_line, right_line = detect_lines_from_previous_lines(warp_binary, left_line, right_line)
cal_curvature_and_pos(left_line, right_line)
if prev_left_line is not None and (abs(left_line.line_base_pos - prev_left_line.line_base_pos) / prev_left_line.line_base_pos) > 0.1:
left_line.detected = False
if prev_right_line is not None and (abs(right_line.line_base_pos - prev_right_line.line_base_pos) / prev_right_line.line_base_pos) > 0.1:
right_line.detected = False
# Create an image to draw the lines on
warp_zero = np.zeros_like(warp_binary).astype(np.uint8)
color_warp = np.dstack((warp_zero, warp_zero, warp_zero))
# Generate x and y values for plotting
ploty = np.linspace(0, img.shape[0]-1, img.shape[0] )
left_fitx = left_line.current_fit[0]*ploty**2 + left_line.current_fit[1]*ploty + left_line.current_fit[2]
right_fitx = right_line.current_fit[0]*ploty**2 + right_line.current_fit[1]*ploty + right_line.current_fit[2]
# Recast the x and y points into usable format for cv2.fillPoly()
pts_left = np.array([np.transpose(np.vstack([left_fitx, ploty]))])
pts_right = np.array([np.flipud(np.transpose(np.vstack([right_fitx, ploty])))])
pts = np.hstack((pts_left, pts_right))
# Draw the lane onto the warped blank image
cv2.fillPoly(warp_image, np.int_([pts]), (0,255, 0))
# Warp the blank back to original image space using inverse perspective matrix (Minv)
newwarp = cv2.warpPerspective(warp_image, M_inv, (img.shape[1], img.shape[0]))
# Combine the result with the original image
result = cv2.addWeighted(undist, 1, newwarp, 0.3, 0)
curverad = round((left_line.radius_of_curvature + right_line.radius_of_curvature) / 2)
curverad_str = 'Curve radius: ' + str(curverad) + 'm'
lane_base_pos = round((left_line.line_base_pos + right_line.line_base_pos) / 2, 4)
if lane_base_pos > 0:
position_str = str(lane_base_pos * 100) + 'cm left of center'
elif lane_base_pos < 0:
position_str = str(lane_base_pos * -100) + 'cm right of center'
else:
position_str = 'right at center'
cv2.putText(result, curverad_str, (20, 70), cv2.FONT_HERSHEY_DUPLEX, 2, (0,255, 0), 2, cv2.LINE_AA)
cv2.putText(result, position_str, (20, 140), cv2.FONT_HERSHEY_DUPLEX, 2, (0,255, 0), 2, cv2.LINE_AA)
return result
fig = plt.figure(figsize = (20, 50))
for i in range(len(test_images)):
test_image = test_images[i]
image = mpimg.imread(test_image)
left_line = None
right_line = None
output_image = draw_lane(image)
ax1 = fig.add_subplot(8,2,(i*2)+1)
ax1.set_title(ntpath.basename(test_image) + ': Original')
ax1.imshow(image)
ax2 = fig.add_subplot(8,2,(i*2)+2)
ax2.set_title('Lane Highlighted')
ax2.imshow(output_image)
Video¶
Import Packages¶
In [14]:
from moviepy.editor import VideoFileClip
from IPython.display import HTML
Basic Video¶
In [15]:
left_line = None
right_line = None
video_output = 'project_video_output.mp4'
video_input = VideoFileClip('project_video.mp4')
processed_video = video_input.fl_image(draw_lane)
%time processed_video.write_videofile(video_output, audio=False)
In [16]:
HTML("""
<video width="960" height="540" controls>
<source src="{0}">
</video>
""".format(video_output))
Out[16]:
Challenge Video¶
In [17]:
left_line = None
right_line = None
challenge_video_output = 'challenge_video_output.mp4'
challenge_video_input = VideoFileClip('challenge_video.mp4')
processed_challenge_video = challenge_video_input.fl_image(draw_lane)
%time processed_challenge_video.write_videofile(challenge_video_output, audio=False)
In [18]:
HTML("""
<video width="960" height="540" controls>
<source src="{0}">
</video>
""".format(challenge_video_output))
Out[18]:
Harder Challenge Video¶
In [19]:
left_line = None
right_line = None
harder_challenge_video_output = 'harder_challenge_video_output.mp4'
harder_challenge_video_input = VideoFileClip('harder_challenge_video.mp4')
processed_harder_challenge_video = harder_challenge_video_input.fl_image(draw_lane)
%time processed_harder_challenge_video.write_videofile(harder_challenge_video_output, audio=False)
In [20]:
HTML("""
<video width="960" height="540" controls>
<source src="{0}">
</video>
""".format(harder_challenge_video_output))
Out[20]: